PRIVATE PENSIONS: COVERAGE AND BENEFIT TRENDS. Alicia H. Munnell and Annika Sundén With the Assistance of Elizabeth Lidstone *

Transcription

1 PRIVATE PENSIONS: COVERAGE AND BENEFIT TRENDS Alicia H. Munnell and Annika Sundén With the Assistance of Elizabeth Lidstone * Prepared for Conversation on Coverage Washington, D.C. July 24-25, 2001 * The authors are at Boston College. They are Peter F. Drucker Professor of Management Sciences at Boston College School of Management and Director of the Center for Retirement Research, Associate Director for Research of the Center for Retirement Research, and Research Associate, respectively. The authors would like to thank Sean Barrett, Mireille Samaan, Mauricio Soto and Catherine Taylor for able research assistance. Many people were very generous in providing data, expertise, and computer code during the preparation of this paper. Alan Gustman, Tom Steinmeier and Courtney Coile provided their computer programs for calculating pension wealth, John Woods at the Social Security Administration (SSA) helped us manipulate the 1993 Pension Supplement to the CPS and reviewed a previous draft, and Patrick Purcell of the Congressional Research Service helped us reconcile our coverage numbers with those he has produced. Craig Copeland, Susan Grad, Alan Gustman, Ken McDonnell, Anna Rappaport, David Raynes, Dallas Salisbury, and Jack Vanderhei reviewed the entire document. Although we tried to respond to all our reviewers, they may still have some disagreements on particular issues. In the end, the opinions and conclusions are solely those of the authors. They should not be construed as necessarily representing the opinions or policy of any of the reviewers or of SSA or any agency of the Federal Government.

2 The U.S. retirement income system has often been described as a three-legged stool. The first leg is the public Social Security system, which covers virtually all workers and provides benefits based on lifetime earnings at 65 (gradually increasing to 67 by 2022), and reduced benefits at 62. The second leg consists of employer-provided pensions, which cover roughly half the work force at any point in time. These tax-subsidized plans are sponsored by private employers, by the federal government for its employees, and by state and local governments for their workers. The third leg was said to consist of individual saving, which occurs in tax-subsidized Individual Retirement Accounts (IRAs) or directly in non-subsidized forms. The three-legged-stool concept probably needs to be updated in two ways. First, individuals currently do little saving outside of employerprovided pensions and most of the money in IRAs is the result of rollovers from employment based plans. Second, health insurance has become an important leg of the stool. Regardless how one describes the retirement income system, employer-provided pensions play an important role in assuring a comfortable retirement. This paper focuses specifically on pensions provided by private, rather than government, employers. It explores who is covered by a private pension plan and who is not, how much retirees receive in pension income, and how pension coverage and pension receipt have changed over time. By necessity, the focus of this survey is narrow. It does not explore all sources of support in retirement. Specifically, the Supplemental Security Income program that provides benefits to those with very low income and virtually no assets ($2,000 for an individual, $3,000 for a couple) is beyond the scope of the current study. The same is true for the nation s major health programs Medicare and Medicaid. Changing employment patterns of older workers are also not addressed herein. The focus of this survey is the trends and patterns of coverage under private sector employer-sponsored plans, and in some cases the relationship of these plans to Social Security.

3 The trends and patterns that emerge from this survey are as follows. After increasing sharply in the post-world War II period, the percentage of the private sector work force covered by an employer-sponsored pension plan at any given point in time has remained around 50 percent since the 1970s. This constancy obscures two major changes, however. First, pension coverage has increased for women and declined for men, primarily reflecting the increased earnings and labor force attachment of women and a decline for men in unionism and employment in large manufacturing firms. Second, a major shift has occurred in the types of plans from defined benefit to defined contribution, reflecting employment trends as well as conversion of plans. Whereas only about half the work force is covered at a given point, pension coverage is more extensive when considered over workers lifetimes and on a household basis. In addition, more rapid vesting means that more plan participants are assured of receiving benefits. On the other hand, pensions are more important for high-income than for lowincome workers. This would not be a problem if Social Security alone allowed those at the low end of the income scale to maintain their living standards in retirement, but it does not. All workers need more than Social Security for a comfortable retirement. The primary challenge for those interested in retirement security is to expand coverage so that more workers have enough income in retirement to avoid sharp drops in their living standards. About one quarter of those without coverage are employed in firms where the employer sponsors a plan; here the task is to make sure that employees have the knowledge and information they need to make a fully informed decision on whether or not to participate. The other three-quarters of those without a pension work for employers who do not sponsor a plan. The task here is to encourage more employers particularly small employers to provide a pension. The search for new incentives and innovative approaches to expand coverage leads us to today s Conversation on Coverage. 2

4 I. A Brief History Private pension plans in the United States date from 1875, but the early plans were financially vulnerable and most were bankrupted in the 1930s by the Great Depression. Contemporary U.S. pension plans, both private and public, are rooted in the search for greater financial security that became part of the national psychology after the onset of the Depression. Although World War II initially consumed much of the nation s resources that might have been directed toward improved provisions for old age, wartime wage controls greatly stimulated the expansion of private plans. Since legal limitations on cash wages impeded the ability of employers to attract and hold workers in the tight civilian labor markets, the War Labor Board attempted to relieve the pressure on management and labor by permitting employers to bid for workers by offering attractive fringe benefits. Pension benefits cost firms little in view of the wartime excess profits tax and the ability to deduct contributions. The growth of new pension plans fell off markedly in the immediate post-war period as employees focused on cash wages to recover ground lost during the period of wage stabilization, but by 1949 pension benefits again became a major issue of labor negotiation (Munnell 1982 and Sass 1997). The main expansion of today s pension system, in both union plans and non-unionized industries, began in the 1950s. The Korean War further stimulated the pension movement as employers once again competed for workers in the face of wage and salary controls and excess profits taxes. The growth in pensions continued in the 1960s. But much of the growth during that decade was due to expansion of employment in firms that already had pension plans as opposed to establishment of new plans. The percentage of the private work force covered by any type of employer-sponsored retirement plan continued to increase until the late 1960s, but since then coverage has remained unchanged. 3

5 II. Trends in Pension Coverage Before discussing pension trends, it is necessary to clarify three distinct ways in which workers can be associated with a plan. The first is that they could work for an employer that sponsors a plan to any of its employees. The second is that they could be covered by a plan, but not be eligible for benefits. And the third is that workers actually participate in the plan. Coverage and participation are not the same, since, for example, a percentage of workers covered in 401(k) plans choose not to participate. Nevertheless, this paper will use the terms coverage and participation interchangeably, except in the discussion of 401(k) plans. Data on pension coverage are provided by two main sources the Current Population Survey (CPS), which is administered jointly by the Bureau of Labor Statistics (BLS) and the Bureau of the Census, and the Employee Benefits Survey (EBS), conducted by the BLS. 1 Although the EBS indicates more coverage than the CPS for comparable populations, the two series provide a relatively consistent picture of pensions in the United States. The CPS interviews roughly 50,000 households each month, primarily about labor force activity, to estimate the unemployment rate. Each March, the survey includes supplemental economic and demographic questions and questions about income and employment during the previous year. Respondents are asked whether any employer for whom they worked in the previous year had a pension or other type of retirement plan for any of its workers. Those who answer yes to this question are asked whether they were included in the plan. Consistent March CPS data are available since The number of participants is also reported in the Internal Revenue Service s Form 5500, and theoretically it is possible to calculate coverage rates by dividing participants by the non-agricultural, nongovernment work force. The problem is that the Form 5500 data double-counts people covered by two plans. That is, if the employer provides both a defined benefit plan and a 401(k), the worker will be counted twice. As a result, estimating participation with the Form 5500 would significantly overstate the percentage of the work force covered by a pension plan. 2 In addition to the March data, the BLS has conducted supplementary surveys on employee benefits in April 1972, in May 1979, 1983, and 1988, and in April Although the BLS has no current plans to repeat those surveys, questions on health and retirement benefits were included in CPS supplements on 4

6 The second source of information on pension participation is the Employee Benefits Survey (EBS). In even-numbered years, the BLS surveys establishments with fewer than 100 employees, and in odd-numbered years the agency surveys medium and large employers. Data for medium and large employers are available for selected years from 1980 to 1997; data for small employers are available since EBS estimates of participation in retirement plans are higher than estimates derived from the CPS. Among full-time workers, the gap between the two surveys is 10 percentage points (Herz et al. 2000). The difference between the two surveys can be attributed to sampling procedures and survey methods (Purcell 2000a). The main advantage of the EBS is that it surveys business establishments and asks human resource professionals a lengthy list of questions about the business s employee benefit plans. The CPS interviews a single respondent in a household and devotes only a fraction of the interview to benefits. Respondents may lack knowledge about their own benefit coverage and particularly about the benefit coverage of other household members for whom they responded (Herz et al. 2000). 4 The disadvantages of the EBS for looking at general trends are 1) the survey interviews small and large employers separately, so that total coverage rates are not available annually and 2) the EBS historical information is limited. For these reasons, the following discussion focuses on trends in the CPS. workers in contingent and alternative employment arrangements conducted in February of 1995, 1997, and In addition, Topical Modules 4 and 7 from the 1996 Panel of the Survey of Income and Program Participation (SIPP) continues to ask some of those questions, as will the 2001 and subsequent panels of that survey. 3 The definition of medium and large establishments changed in Until then, the BLS included establishments that employed at least 50, 100, or 250 workers depending on the industry. Thus, in mining, construction, retail trade, some manufacturing and some transportation, the BLS approached only firms with 250 employees or more. In accounting, auditing, and bookkeeping, the minimum establishment size was 50 employees. In 1988, the BLS decided to focus on all establishments with a 100 or more employees, and this change increased the size of the sample by 50 percent (Mitchell 2000). 4 The CPS and EBS also differ in a few other ways. The EBS participation measure includes workers who have not satisfied their employer s length of service requirement for participation in the retirement plan, while the CPS measure excludes these individuals. Another difference between the two surveys is that the CPS includes self-employed incorporated workers as private workers while the EBS does not. Selfemployed incorporated workers are a relatively small portion of the work force (2.8 percent in 1999), and tend to have low rates of pension coverage (27 percent of workers aged participated in a pension plan in 1999). 5

7 The CPS can yield a wide range of numbers for pension coverage, depending on how one defines coverage and the relevant population. This is shown clearly in Figure 1 where the population moves gradually from private and government full-time full-year workers aged whose employer sponsors a pension plan to all private sector workers who actually participate in a plan. Including government workers, restricting the relevant labor force substantially, and using employer sponsorship as the relevant criteria indicates that 68 percent of the population had at least the potential for pension protection in At the other extreme, focusing only on private sector workers and eliminating the age and full-time constraint shows that 42 percent of private sector workers participated in a pension. What s the relevant number? First, this discussion focuses on the private sector, so government workers should be excluded. In fact, government workers enjoy a high level of pension coverage with 87 percent of full-time full-year workers aged participating in a pension plan in Second, the real concern is not simply whether the employer offers a plan to any of its workers but whether the worker actually participates, so participation, as opposed to sponsorship, is the important criteria. Third, part-time and part-year workers should probably be included since they are frequently regular participants in the work force. 5 Fourth, young workers that is, those under 25 generally are more concerned about establishing themselves in their careers than accumulating assets for retirement, so their pension coverage is not a major concern. 6 That leaves the second line from the bottom in Figure 1, which shows that only about 50 percent of private sector workers age participated in an employer-sponsored pension in But one could argue that all the numbers are relevant, depending on 5 The IRS non-discrimination provisions already require inclusion of workers with 1000 hours within a 12- month period. Some state plans, such as California State Teachers Retirement System (CalSTERS), go beyond this minimum and offer pension plans for part-time and seasonal employees. 6 Under the Employee Retirement Income Security Act (ERISA) of 1974, full-time employees age 25 or older must be granted participant status after completing one year of service. The Retirement Equity Act of 1984 subsequently amended the participation rules, however, and for most plans lowered the participation requirement to age 21 (McGill et. al 1996). 7 Participation for those age is exactly the same in 1998 and 1999 as for those aged and within two tenths of a percentage point for the 1990s. In the late 1970s and 1980s, participation for those was slightly (less than 1 percentage point) lower, but this small discrepancy disappeared as the retirement age fell. The reason is that a disproportionate share of those age withdrawing from the labor force 6

8 the purpose a narrow definition of the work force to determine the success of ERISA, a broader definition of the work force to access the possibilities for expansion of coverage, and employer sponsorship to look at potential participation. While the level of pension participation depends on definitions, the trend over time does not. Regardless of how the relevant population is defined, pension participation is just about where it was in In each case, participation dropped between 1979 and 1988 and then rebounded between 1988 and In both 1979 and 1999, only 50 percent of non-agricultural wage and salary workers in the private sector aged participated in a pension plan, even though 1979 was the end of a decade of stagnation and 1999 was the height of the longest expansion in the post-war period. In fact, a simple correlation between pension participation and the unemployment rate suggests that much of the rebound is the result of the strong economic growth and high levels of employment in the late 1980s and 1990s. Coverage by Sex, Earnings, and Race Although the overall participation rate remained virtually unchanged between 1979 and 1999, that stability was the result of offsetting changes for males and females. Figure 2 shows that pension coverage declined for all male workers except those in the highest earning quintile (Panel A). In contrast, participation for women increased at all earnings levels (Panel B). The drop in male participation rates was caused by declines in unionism and employment at large manufacturing firms, and by the rapid growth in 401(k) plans that made employee participation in pensions voluntary. 8 Among women, the growth in pension participation was largely the result of improved earnings and an increase in fulltime work and to a lesser extent increased unionism and employment at large firms (Even and Macpherson 1999). The remaining differential between coverage patterns for were covered by a pension plan, which reduced the participation rate for this older group. This reduction in coverage among older workers eventually made the participation rate for those aged equal to that of those aged Even and Macpherson (1994) showed that the growth of 401(k) plans caused participation rates to drop most for young and less educated workers. 7

9 men and woman can be explained by their different work patterns, since pension coverage among women who work full-time, full-year is virtually identical to the coverage rates for men (Copeland 2000). Figure 2 shows that participation is closely correlated with earnings levels. In the top quintile, about 70 percent of workers both male and female participate in pensions; in the bottom quintile, that figure drops to about 20 percent for men and 10 percent for women. Earnings also appear to be more important than race in explaining pension participation. Overall, pension participation for minority groups is lower than for whites. In 1999, pension participation was 44 percent for blacks and 31 percent for Hispanics, compared to 54 percent for non-hispanic whites. However, as shown in Figure 3, when examining pension participation by earnings quintiles, the picture for whites and blacks looks very similar. Pension participation in the top quintile is about 75 percent for both groups and it drops to about 15 percent in the bottom quintile. Hispanics, on the other hand, have lower pension participation rates in all earnings quintiles in the top quintile about 65 percent have a pension, and this share drops to 10 percent in the bottom quintile. On balance, however, earnings appear to be a more important factor than race in determining whether a worker participates in a pension plan. 9 This examination of pension coverage by various groups makes clear once again that about half the work force is not covered by a pension. The Uncovered Firm Has a Plan Of those not covered by a pension plan, roughly one quarter work for an employer with a plan and three-quarters are employed in a firm without a plan. 10 The Internal Revenue 9 Data by race and sex show a very similar pattern. Both male and female non-hispanic whites and blacks have roughly the same rate of pension participation within each earnings quintile; the rates for Hispanics are consistently lower for both males and females. 10 The March 2000 CPS shows that 22.5 percent of those aged without a pension work in firms with a plan and 77.5 percent are employed by companies without a pension. 8

10 Service (IRS) s non-discrimination provisions allow firms to exclude employees under age 21 or with less than one year of employment with the firm. Since a year of service is defined as 1000 hours during a 12-month period, many part-time and seasonal workers never qualify to participate in the plan (Halperin and Munnell 2001). In addition to the exclusion for age and service, the firm is also allowed to exclude up to 30 percent of the remaining rank-and-file workers from the plan, although other non-discrimination tests make an exclusion of this magnitude unlikely. 11 As shown in Figure 4, more than half of those excluded from firm-sponsored pension plans do not meet the age and service requirements or do not work enough to qualify for the plan, and another 6 percent were excluded because their job was not eligible for pension coverage. An increasingly important reason for not participating is the employee choosing not to contribute, presumably in 401(k) plans. Researchers have explored the reasons why workers participate or do not participate in 401(k) plans. Part of the behavior can be explained by the characteristics of workers themselves. A series of studies using CPS data shows that participation and the level of contributions are related positively to income, age, education and job tenure (Andrews 1992, EBRI 1994, and Bassett, Fleming, and Rodriguez 1998). A recent study using the 1998 Survey of Consumer Finances also documents the importance of an individual s planning horizon in the participation and contribution decision (Munnell, Sunden and Taylor 2000). This finding reinforces conclusions from Bernheim (1998) and Clark and Schieber (1998) based on plan data indicating that employer provided information can be very important in changing employees attitudes about the necessity of saving for retirement. Plan structure and characteristics also are important. If employers offer a match, workers are more likely to contribute since the match provides a large initial return that supplements the advantage of deferral. Researchers have not reached a consensus, however, as to whether employees respond to the level of the match rate once a positive 11 Although section 410(b) allows the sponsor to exclude up to 30 percent of other workers, the average deferred percentage (ADP) and the actual contribution percentage (ACP) tests make this difficult. 9

11 match is provided. 12 If the 401(k) plan supplements a defined benefit plan, workers are less likely to participate. The research suggests that both participation and contributions are negatively related to the presence and generosity of a defined benefit plan. Finally, workers are more likely to participate and contribute if they can gain access to their funds through borrowing. Studies have shown that individuals behavior often reflects a surprising amount of inertia and that if employees are automatically enrolled in a plan, they are more likely to participate than if they had to opt in (Madrian and Shea 2000). In 1998 and 2000, the IRS issued regulations that permit employers to enroll employees automatically in 401(k), 403(b), and 457 pension plans. Benefit consultants estimate that since the IRS issued its first regulation, 7 to 10 percent of 401(k) plan sponsors have adopted automatic enrollment (Jacobious 2000 and Purcell 2001). A study by Hewitt Associates found that only 4 percent of people automatically enrolled in a 401(k) plan subsequently opted not to participate. Thus, it is not surprising that, according to a recent study of 10 companies, average participation rates rose from 76 percent to 93 percent after enrollment was adopted (Purcell 2001). Uncovered Firm Does Not Have a Plan The majority of uncovered workers are employed in firms without a pension plan, and the existence of a pension plan varies sharply by size of firm. The CPS shows that 32 percent of employees aged in firms with less than 100 workers participate in a pension, compared to 72 percent for firms with more than 100 employees. Although the level of coverage in the EBS is higher for both small and large employers, the differential 12 Kusko, Poterba and Wilcox (1998) find little change in either participation or contributions in response to large changes over time in matching provisions. Bassett, Fleming, and Rodriguez (1998) find no evidence that participation rises with the match rate. Papke (1995) shows that participation increases with the level of the match rate, with smaller marginal effects at higher match rates, and that contributions increase markedly as the employer moves from a zero to a positive match rate, with a negative effect at very high match rates. Papke and Poterba (1995) conclude that participation increases with the match rate but find no significant effect on contributions. Clark et al (2000) find a positive effect of the match rate on participation and a negative effect on contributions. VanDerhei, Copeland, and Quick (2000) confirm the finding that the match rate has a negative effect on contributions. Munnell, Sunden, and Taylor (2000) find that the level of the employer match is not particularly important for contributions given the match exists. 10

12 by size in the two surveys is roughly similar (Figure 5). The CPS provides a further breakdown by firm size. It shows that among private sector workers age the participation rate increases from 23 percent in firms with less than 25 employees, to 43 percent in firms with 25 to 99 employees and to 64 percent for firms with 100 or more employees (Figure 6). As reasons for not providing coverage, small employers frequently mention high employee turnover and the preference of their employees for cash wages. They also cite uncertainty about future earnings and the newness of the business. Table 1, taken from a survey of small employers by the Employee Benefits Research Institute (2001), documents the relative importance of these various factors. Note that for the entire small business sample only 12 percent say that a major reason for not providing a pension is that it costs too much to set up and administer a plan. Even defining cost broadly to include required contributions too expensive and too many government regulations brings the total to only 26 percent. Surprisingly, those firms with employees view cost as a more important factor than very small firms (those with 5-20 employees). Twenty-two percent of firms with employees responded that the expense of setting up and administering a plan was the major reason for not having a pension, and 31 percent cited a broad definition of costs as a major factor. Nevertheless, the survey suggest that for small firms cost is an important but not dominant consideration. The low level of pension coverage in small firms is an important policy concern since about one-third percent of full-time workers are employed in firms with less than a hundred workers (Copeland 2001b). In an effort to make it easier for firms to establish and maintain plans for their employees, the federal government has passed several pieces of legislation over the years to ease financial and reporting requirements. 13 But, as noted 13 The Revenue Act of 1978 authorized a defined contribution plan for firms with fewer than 100 employees called the Simplified Employee Pension (SEP). The Tax Reform Act of 1986 allowed workers in firms with less than 25 employees to contribute to a SEP on a tax-deferred basis through salary reduction (SARSEP). The Small Business Job Protection Act of 1996 authorized another type of defined contribution plan called Savings Incentive Match Plans for Employees of Small Employers (SIMPLE). SIMPLE plans generally replaced SEPs, although firms with under 100 employees could continue to establish SEPs funded exclusively by employer contributions (Purcell 2000a). 11

13 above, while cost and administrative issues matter, they are not the primary reason for low sponsorship among small employers. Perhaps that is the reason that despite this legislation, the discrepancy in coverage between large and small firms remains. Alternatively, small business owners may not have had time to understand the new plans particularly those included in the 1996 legislation. In addition to employer size, the GAO (2000) provides a useful summary of the major characteristics of those not covered by a pension plan. It finds that 85 percent of those in the labor force who work and lack pension coverage have low income, do not work full time, work for a small firm, and/or are relatively young. Figure 7 shows the share of employees with each of these characteristics that did not have coverage. Lifetime Pension Coverage The CPS March data on pension coverage apply only to individual workers at any given point in time. Over a lifetime and on a household rather than an individual basis, coverage rates are somewhat higher. This can be seen by looking at the 1993 CPS Pension Supplement, the last detailed survey of pension characteristics, at the Health and Retirement Study (HRS), and at the Survey of Consumer Finances (SCF). The HRS provides information on the income and wealth holdings for a nationally representative sample of 7,607 families who had at least one member born between 1931 and That is, during the first wave of interviews in 1992 respondents were between the ages of 51 and 61. The SCF is a triennial survey of a nationally representative sample of the entire population sponsored by the Federal Reserve Board in cooperation with Statistics of Income of the Department of the Treasury. The SCF collects detailed information on approximately 4,000 households' assets, liabilities, and demographic characteristics as well as on pension coverage, participation, and pension plan characteristics such as contribution levels. The survey over-samples wealthy households to provide reliable estimates of highly concentrated assets. 12

14 The 1993 CPS Pension Supplement asks questions about pension coverage not only at the worker s current job, but also at the worker s current second job and at previous jobs. Table 2 shows the percentage of men and women who have coverage under a private employer-provided plan sometime during their life. Pension coverage rises with age, so 66 percent of male and 60 percent of female workers aged have some pension coverage. To get coverage information on a household basis, it is necessary to go to the HRS or the SCF. Figure 8 presents lifetime pension coverage for households aged by income quintiles for the two surveys. Both the HRS and SCF show that approximately 65 percent of households had some sort of pension coverage in However, the figure also shows that pension coverage is much more extensive for those households in the top income quintiles. Both the HRS and the SCF indicate a marked drop off of pension coverage from above 80 percent in the top two quintiles to 30 percent for the bottom quintile. Thus, the CPS annual statistics on pension participation understate the proportion of the population that will be eligible for pension benefits in retirement. But looking at coverage even on a lifetime basis does not change the fact that pension coverage is significantly greater for those with high income than for those at the low end of the income distribution. III. A Shift to Defined Contribution and Cash Balance Plans Although the percentage of the work force covered by a pension plan has remained virtually unchanged since the 1970s, the nature of pension coverage has moved sharply from defined benefit plans to defined contribution plans. Defined benefit plans generally provide retirement benefits based on a percentage of final salary for each year of service, 14 Although more recent SCF data are available for 1995 and 1998, the 1992 data are presented for comparability with the 1992 HRS. SCF data for 1995 and 1998 provide a similar picture. 13

15 and pay the benefits in the form of a lifetime annuity. 15 For example, a worker with a final salary of $40,000 might receive 1.5 percent a year for 30 years of service, producing an annual pension of $18,000. The employer pre-funds these benefits by making pre-tax contributions into a pension fund; employees typically do not contribute. The employer holds the assets in trust, directs the investments, and bears the risk. Benefits are insured up to specified limits by the Pension Benefit Guaranty Corporation (PBGC). 16 The defined benefit form of pension made it easy for employers at the time that they established their plans in the 1950s to grant retroactive credits for older workers for whom they had made little or no contributions. In contrast to defined benefit plans, defined contribution plans are like savings accounts. Generally the employer, and often the employee, contributes a specified dollar amount or percentage of earnings into the account. These contributions are invested, usually at the direction of the employee, in mutual funds consisting of stocks and bonds. When the worker retires, the balance in the account determines the retirement benefit. A Shift from Defined Benefit to Defined Contribution Plans Information on the shift from defined benefit to defined contribution plan coverage and participation comes from two sources: the Labor Department s Employee Benefit Survey discussed above and the IRS s Form Almost all sponsors of pensions subject to ERISA must file Form 5500 annually with the IRS, and the Department of Labor periodically publishes summaries of the data. 17 Figure 9 presents Form 5500 data for 1975, the first year the forms were filed, and 1997, the most recent data available. Over this period, assets in defined contribution plans rose from 28 percent to about 50 percent of total pension assets; benefits paid by defined contribution plans rose from 32 percent to 58 percent of total benefits; active participants in defined contribution plans increased 15 Lifetime annuities are paid only to those who have remained with the employer long enough to have a benefit worth more than $5,000 in present value. 16 The PBGC monthly guarantee limit in 2001 is $3,392 at age 65, and declines to $1,546 at age Small employers with SIMPLE and SEP plans are not required to file the Form Also not required to file the Form 5500 are sponsors of pension plans with total assets of $100,000 or less at the end of every plan year beginning on or after January 1,

16 from 29 percent to 68 percent of total participants; and contributions to defined contribution plans increased from 35 percent to 83 percent of total contributions. Within the defined contribution world, the fastest growing type of plan is the 401(k). Similar plans had existed for decades but they were clearly authorized in the Revenue Act of 1978, which ended ambiguity surrounding their status. They became popular and spread after the IRS issued clarifying regulations in The defining characteristics of 401(k) plans are that participation in the plan is voluntary and that the employee as well as the employer can make pre-tax contributions to the plan. These characteristics shift a substantial portion of the burden for providing for retirement to the employee; the employee decides whether or not to participate, how much to contribute, and how to invest the assets. In addition, workers have access to 401(k) plan funds before retirement, adding another element of individual responsibility. Despite the fact that employees bear much of the risk in 401(k) plans, these plans have grown enormously for a number of reasons. They are more appealing to a younger, more mobile work force. For these workers, greater portability clearly outweighs the predictability of benefits for the career employee under a defined benefit plan. Workers get statements several times a year and can see their balances grow, which makes defined contribution benefits seem more tangible. From the employer s perspective, 401(k) plans may be less costly to operate than defined benefit plans. They do not require employer contributions, although most employers do contribute to these plans. And, they are fully funded by definition, eliminating the work associated with funding requirements and pension insurance. Portability in some cases eliminates the need for employers to keep track of pensions for departed employees. 18 As shown in Figure 10, between 1984 (the first year separate data are available for 401(k) plans) and 1997, all dimensions of 401(k) plans assets, benefits, participants, and contributions have increased from 25 to 35 percent of total defined contribution plans to 70 to 80 percent. 18 Many employees still leave accounts behind requiring the employer to pay some record keeping fees. This administrative burden may help explain the trend toward increased lump sum payments in defined benefit plans. 15

17 Given their popularity and growth, one would have thought that the introduction of 401(k) plans should have boosted pension plan coverage in the United States. But, as noted above, overall pension coverage has remained virtually unchanged. This means that the enormous expansion of defined contribution plans, especially 401(k)-type plans, has produced a sharp decline in the percent of the work force covered under traditional defined benefit plans. This decline reflects shifts in employment from manufacturing to service industries and to a lesser extent employers substituting defined contribution for defined benefit plans. Researchers attribute about half of the decline in defined benefit coverage can be attributed to employment shifts and half to substitution (Ippolito and Thompson 2000 and Gustman and Steinmeier 1992). Most of the employer substitution of plans occurred in the late 1980s and 1990s. Before 1986, a large number of 401(k) plans were converted thrift plans, which had typically supplemented defined benefit plans but allowed only after-tax employee contributions. Engen, Gale and Scholz (1996) contend that even by 1991, a large portion of 401(k) assets was still in plans originating before Over the period , however, Papke (1999) found that about 20 percent of ongoing sponsors dropped defined benefit plans entirely in favor of defined contribution plans, and that they replaced traditional defined contribution plans with 401(k) plans. Papke may overstate the rate of substitution slightly since she does not take into account plan mergers and changes in plan identification numbers. 19 Nevertheless, substitution as well as increased employment in the service industry explain why coverage has not increased. Data on households from the SCF presented in Figure 11 confirm the shift in coverage rates from defined benefit to defined contribution plans. The share of households with only a defined benefit plan dropped from about 40 percent to 20 percent between 1992 and 1998, while the share that rely solely on a defined contribution plan increased from 37 percent to 57 percent (VanDerhei and Copeland 2001). 19 Ippolito and Thompson (2000) points out that the number of plan sponsors with defined benefit plans that remain over the time period is underestimated if the sample is not adjusted for plan mergers and changes in plan names. Applying Ippolito and Thompson's adjustment to Papke's analysis indicates that the share of firms that drop their defined benefit plan is 14 percent rather than 20 percent. 16

18 The household data also indicate that 22 percent of households have dual coverage and that this share has held steady through the 1990s. 20 The constant share of households with dual coverage is consistent with data from the Form The proportion of private sector workers that were covered by both a defined benefit and defined contribution plan increased from 10 percent to 17 percent between 1977 and 1983 but has been constant since then. The slightly higher level of dual coverage found in the SCF probably reflects that it measures coverage for households while Form 5500 measures coverage for individuals. A Shift of Defined Benefit Plans to Cash Balance Plans In addition to the shift in pension coverage from defined benefit to defined contribution plans, some employers have converted their pensions to hybrid plans that have both defined-benefit and defined-contribution characteristics. The most popular of the hybrids are the so-called cash balance plans. Bank America created the first cash balance plan in 1985, but few employers followed suit until the late 1990s. Legally, cash balance plans are defined benefit plans where the employers prefund contributions, own the assets, select the investments, and bear the risk. And, like other defined benefit plans, the PBGC insures the benefits. To the employee, however, cash balance plans look very much like a defined contribution plan. Contributions made for the employees are recorded in a notional account. The employees receive regular statements showing the balance in their notional account, and the benefits tend to accrue as a constant percentage of compensation. At separation, the employee can withdraw the balance, which is usually greater than they would get under a traditional defined benefit plan. The plans are attractive to employees because they provide visible and portable benefits like a defined contribution plan, while securing accrual and government insurance like defined benefit plans. Moreover, for the employer to switch from a traditional defined benefit plan to a 20 Due to employment growth the number of active participants in pension plans have increased since The absolute number of active participants in defined benefit plans has remained fairly constant while all new growth in employment has been in industries that offer defined contribution plans. As a result, the share of workers covered by defined benefit plans has decreased. 17

19 cash balance plan requires only a plan amendment, whereas switching to a defined contribution plan requires terminating the defined benefit plan, which might trigger a reversion excise tax of either 20 or 50 percent (Gale, Papke, and Vanderhei 2001). 21 The most recent official statistics report that only 6 percent of full-time employees at medium and large private establishments had a cash account benefit formula in But surveys suggest that significant conversion has occurred since then. About 16 percent of pension plans among the Fortune 100 were cash balance plans in 1998, and more generally cash balance plans may have increased from 5 percent to 12 percent of all defined benefit plan between 1998 and 2000 (Elliot and Moore 2000). One interesting aspect of cash balance plans is that they provide lump-sum payments at separation. So the growth of lump sums comes not only from the dramatic expansion of defined contribution plans but also from the changing nature of defined benefit plans. Assuming only modest growth in cash balance plans, lump-sum pension benefits should exceed annuity payments from defined benefit plans for those retiring in 2010 (VanDerhei and Copeland 2001). A Comment on the Economics of Pension Coverage One comment on the economics of pension coverage is necessary before proceeding. The shift from defined benefit to defined contribution plans has no obvious implications for whether employers are providing more or less for their workers. Economists contend that the introduction of a pension implies a reduction in cash compensation, even if it takes some time for the adjustment to occur. 22 That is, employers decide on some total 21 Currently the Form 5500 does not provide enough information to identify cash balance plans, but the IRS included additional questions in 1999 so that it will soon be possible to determine the extent to which cash balance plans have spread. 22 The potential trade-off between higher pensions and lower wages may be somewhat more complicated than a simple one-for-one offset. For example, by reducing employee turnover or shirking or by facilitating retirement of less productive workers, the introduction of a deferred compensation arrangement might increase productivity and thus make employers willing to increase total compensation. Alternatively, because of the favorable tax treatment of pensions, the employer has somewhat more money available than if it paid all compensation in cash; the firm could retain this windfall as profit or share some of it with employees, thereby raising total compensation. Even if the trade-off between pensions and wages occurs in the aggregate, it may not happen on a person-for-person basis. For example, raising pensions for minimum 18

20 compensation package that they want to pay their employees and then divide that package between cash wages and fringe benefits. An increase in pension benefits implies a reduction in other benefits or cut in wages and vice versa. Thus, from an economic perspective, the question of who is making the contribution to the plan is less important than it would seem, since economists believe that in the end all pension costs come out of the compensation of the worker. IV. Pensions as a Source of Retirement Income Despite the constancy of the coverage figures, pension benefits from private sector plans have become an increasingly important source of retirement income (Table 3). Part of this increase is the natural result of the expansion and maturation of private plans. That is, coverage expanded significantly after World War II, and as those cohorts retired an increasing number of retirees qualified for private pension benefits. At the same time, vesting has improved considerably. ERISA defined 10 years as the maximum for cliff vesting, and the Tax Reform Act of 1986 required single-employer plans to convert to 5- year vesting. As a result, the EBS reports that in percent of full-time participants in plans sponsored by medium and large employers found their benefits vested after 5 years (Mitchell 2000). With shorter vesting more workers earned assured pensions. Figure 12 shows that the percentage of participants in private sector plans with vested benefits increased from 47 percent in 1979 to 65 percent in 1988 to 84 percent in 1993 (EBRI 1997). Although the CPS provides very useful historical data, it is also interesting to look at the HRS. The two surveys provide a somewhat different picture of the importance of pension income for retirees. Much of the recent confidence in the adequacy of private pensions from the numbers reported in the HRS. wage workers cannot lead to a reduction in wages, because the employer cannot reduce wages below this level. Similarly, plans must cover at least some lower-paid employees to qualify for special tax treatment. But those employees may have no interest in saving for retirement, so plan sponsors may have to use some of the tax benefits to increase the total compensation of "reluctant savers," those who do not much value this deferred compensation. 19

21 Pension Wealth in the HRS and SCF Since the HRS sample was aged 51 to 61 during the first wave in 1992, they were generally not retired and receiving benefits. Nevertheless, the survey asks many questions about expected pension benefits. Moreover, the data include highly detailed employer-provided plan descriptions for pensions covering many workers in the sample, which allows more accurate estimation of pension values than is possible with selfreported data. Social Security records are also available for 70 percent of the sample, which allows accurate estimation of lifetime earnings for each individual and avoids the potential underestimation of expected Social Security benefits that occurs when selfreported data are used. HRS respondents also provide comprehensive information about assets other than pensions and Social Security. The survey deals with the typical high non-response rates when respondents are asked about the value of their stock holdings, IRA balances, or other narrow asset categories by following an open-ended question about value with a series of bracketed questions. From these data, researchers have constructed the wealth holdings of HRS participants in 1992 (Gustman, Mitchell, Samwick, and Steinmeier 2000 and Gustman and Steinmeier 1999b). 23 For defined benefit plans, they use the employer-provided plan descriptions when available to calculate projected benefits and then discount these amounts to estimate pension wealth. When a plan match is not available, they impute a plan based on a series of industry and employment variables. To get pension values as of 1992, they pro-rate wealth on the basis of work to date. That is, if a person age 57 will qualify for an early retirement benefit at age 62 and has been with the firm for 15 years, three-quarters of the eventual accumulation comes from work to date. This pro-rating of pension wealth puts it on an equal footing with other wealth reported. They calculate Social Security wealth in a similar fashion, using the HRS Social Security earnings histories when available and respondent information on current and past earnings when necessary, and 23 Gustman and Steinmeier graciously provided us 8,000 lines of computer code for calculating wealth and other variables in the HRS. 20